[GRASSLIST:7821] Re: r.cost execution time

[Ellen De Beuckeleer]

PROBLEM SOLVED! my stupid newbe mistake :(

My cost layer contained some negative values. That's why the r.cost got 
into an infinite loop.  I hope that by writing this here another newbe 
might not loose so much time...


Ellen De Beuckeleer wrote:

>
> thanx for replying. I tryed to run the r.cost on a smaller map and 
> with percent_memory=100. Moreover, I did run it with the -v option. It 
> still has the problem of getting stuck in the "finding cost path" part 
> (the percentage still raises above 100%). If I do the exact same thing 
> in via the GUI, the green status bar first fills up slowely and once 
> filled completely it starts again and again and again... but then it 
> fills a lot faster. I never get an output map. Can this be because 
> there is something wrong with my command line?
>
> my region is now:
>
> projection: 0 (x,y)
> zone:       0
> north:      -5864
> south:      -6085
> west:       -9543
> east:       -9365
> nsres:      1
> ewres:      1
> rows:       221
> cols:       178
>
> I used start coordinates: -9533, -6023 for my last run
>
> Ellen
>
> Glynn Clements wrote:
>
>> Ellen De Beuckeleer wrote:
>>
>>  
>>
>>> the problem is not solved. My r.cost is still running :(.
>>>
>>> Any suggestions?
>>>   
>>
>>
>> r.cost performs extremely badly on large maps where only a fraction is
>> held in memory. Reducing the region resolution or increasing the
>> percent_memory= option will get around this.
>>
>> If the region is actually 640x480, you should have no trouble with
>> percent_memory=100.
>>
>>  
>>